A large body of international literature has documented a correlation between non-traditional family structure and poorer child outcomes, yet researchers continue to disagree as to whether the association represents a true causal effect. This article extends this literature by employing propensity score matching using the first wave of data from the Growing up in Ireland child cohort study. We argue that the Irish case is of particular interest given the highly selective nature of non-marriage. We find that, on average, non-marriage has negative effects on a child educational development at age 9 but the effects are smaller in relation to health outcomes and the child's self-concept. However, selection effects account for a non-trivial proportion of the differences in child outcomes across lone-mother and cohabiting families although hidden bias remains an important issue. This has important implications for policies which promote marriage as the key to child development as it appears that much of the benefits of marriage are not related to marriage per se but to the socio-economic background of mothers.